The Montreal Cognitive Assessment After Omission of Hearing‐Dependent Subtests: Psychometrics and Clinical Recommendations

Abstract

Objectives Hearing loss (HL) is the third most common chronic health condition in older adults, yet it is often undiagnosed and/or untreated. Given the association between HL and cognitive impairment, it is expected that many people undergoing cognitive screening may have HL. The Montreal Cognitive Assessment (MoCA) is a brief screening test that assesses a wide range of cognitive functions sensitive to Alzheimer's disease (AD) and mild cognitive impairment (MCI). Although MoCA items were carefully designed to be sensitive to deficits in MCI, they were not designed to take sensory declines into account. In the current investigation, we examined the MoCA's psychometric properties following omission of subtests primarily dependent on hearing status (memory, digit span, attention to letters, and sentence repetition). Design Cross‐sectional analytic design (retrospective analysis). Setting We used the original MoCA validation study data.4 Participants Groups consisted of healthy controls (N = 90), subjects with MCI (N = 94), and subjects with mild AD (N = 93). Measurements We assessed sensitivity and specificity using absolute and proportional cutoff score adjustments. We developed receiver operating characteristics curves to determine the best cutoff values for both MCI and AD patients using different combinations of auditory subtest omissions. Results Compared with the original MoCA (MCI sensitivity = 90%; specificity = 87%), MCI sensitivity was substantially reduced (absolute scoring = 43%; proportional scoring = 56%) when all auditory subtests were omitted, with the biggest contribution to the reduction coming from the delayed recall subtest. Excluding three subtests and maintaining the delayed recall had no effect on MCI sensitivity but reduced specificity (sensitivity = 94%, specificity: 71% using proportional scoring). AD sensitivity, in contrast, was not strongly influenced by our manipulation and remained relatively high through all three subtest omission combinations. Conclusion The current study highlights the contribution of hearing‐dependent subtests on the sensitivity and specificity of the MoCA. Clinical recommendations related to these findings are discussed. J Am Geriatr Soc 67:1689–1694, 2019

The Effects of Age-Related Hearing Loss on the Brain and Cognitive Function

“…Because listening with auditory impairment is effortful, older adults with hearing loss may perform worse on these auditorybased cognitive assessments because more cognitive resources are directed towards listening, leaving fewer resources available for the cognitive processing required to perform adequately. The hearing-dependant subtests within tests of cognitive function may significantly affect their sensitivity and specificity as a screening tool [82]. Research shows that omitting the hearing-dependant subtests in one example of these cognitive tests (the Montreal cognitive assessment) reduces the sensitivity in diagnosing mild cognitive impairment; this points at the potential consequences of testing individuals with untreated hearing loss, or testing in a noisy environment, on the accuracy of the these cognitive screening measures [82].…”

Performance on the standard and hearing‐impaired Montreal Cognitive Assessment in cochlear implant users

“…Though it is important to note that HI individuals underperform on the delayed recall task, removing these test items may decrease test sensitivity. 20 This is especially important given that performance on the delayed recall task is the most predictive of MCI or dementia. 21 A second suggestion has been to amplify test material during administration.…”

“…Hillyer et al showed similar results in participants with cochlear implants (CIs). Though it is important to note that HI individuals underperform on the delayed recall task, removing these test items may decrease test sensitivity . This is especially important given that performance on the delayed recall task is the most predictive of MCI or dementia …”

Impact of Hearing Loss on Geriatric Assessment

“…In contrast, sensitivity to detect AD was still between 87% and 100% in comparison to 100% in the original version. 70 Assessment tools that are fully independent of verbal instructions or responses such as the Raven`s Progressive Matrices, which is commonly used for the evaluation of the general intelligence 71 may be another choice. It`s usability as a non-verbal and language-independent test instrument has been proven.…”